In this nested case-control study in the MACS, we used DXA, quantitative CT, and simple anthropometry to compare regional body composition in HIV-infected men with and without clinical evidence of lipodystrophy to HIV-uninfected control subjects. VAT was similar in all 3 groups despite marked differences in BMI, and peripheral lipoatrophy was accentuated in HIV-infected men, regardless of clinical evidence of lipodystrophy, compared to HIV-uninfected men. Finally, over 6 years of follow-up waist circumference increased more rapidly in HIV-infected men who had clinical evidence of lipodystrophy, compared to the HIV-infected men without lipodystrophy and HIV-uninfected men, whereas rate of change in hip and thigh circumference did not differ by group.
While multiple studies have shown that HIV-infected patients with lipodystrophy have more VAT than HIV-infected patients without body composition changes [14, 15], relatively few studies have compared VAT in HIV-infected and HIV-uninfected populations [5, 16]. One of the difficulties in these comparisons is that BMI tends to be higher in HIV-uninfected populations compared to otherwise similar HIV-infected populations, which may be attributable to differences in energy expenditure[17, 18], lipoatrophy, and/or lower lean body mass as a result of chronic HIV infection[19]. In general, VAT tends to be higher in a population with a higher BMI, because of differences in overall adiposity. However, because of lipoatrophy and possibly latent sarcopenia, VAT relative to BMI may be magnified in HIV-infected subjects, posing additional challenges in understanding differences in VAT between HIV-infected and -uninfected subjects.
Cross-sectional studies have taken different approaches to this problem. The FRAM study, the largest such study comparing body composition in men and women with and without HIV infection, reported similar amounts of VAT in 926 HIV-infected and 258 HIV-uninfected subjects[20]. Body mass index, however, was significantly higher in the HIV-uninfected subjects (27.4 ± 5.2 kg/m2 vs. 25.1 ± 4.4 kg/m2, p < 0.0001). In some FRAM analyses[5], the differences in body size were accounted for by adjusting for height or lean body mass measured by MRI. In these analyses, differences in VAT by HIV-status were gender-dependent: HIV-infected women without clinical lipoatrophy had more VAT than HIV-infected women with lipoatrophy or HIV-negative controls[6], while HIV-infected men without clinical lipoatrophy tended to have more VAT than HIV-infected men with lipoatrophy, but less than HIV-uninfected controls[5]. Adjustment for body size did not change these relationships. In another cross-sectional study, Joy and colleagues compared regional fat composition, including VAT, in 306 HIV-infected subjects (70% of whom were categorized as having lipodystrophy), and 107 HIV-uninfected controls[16]. To account for the differences in BMI between the two groups, the authors stratified their analysis by BMI category, arguing that, "weight itself may influence the amount of adipose tissue present". In this analysis, both normal weight and overweight HIV-infected men and women had more VAT than gender-matched, HIV-uninfected controls.
In the present study, as in the MACS as a whole [11], HIV-infected men had lower BMIs than HIV-uninfected men. Nevertheless, VAT was similar between the HIV-infected and uninfected groups. To understand the extent to which the similar VAT levels were confounded by the marked differences in body mass, we adjusted for lean body mass as was done in the FRAM study, and this did not alter the lack of difference between the groups. We also adjusted for BMI and found that the differences between HIV-infected men with lipodystrophy and the other two groups were magnified, the largest difference being that between the two HIV-infected groups. In addition, men with clinical evidence of lipodystrophy tended to have higher VAT than the HIV-uninfected control men after adjustment for BMI.
Consistent with the report by Joy et al, which used stratification, our findings suggest that HIV-infected men with clinical lipodystrophy tend to have more VAT for a given BMI than HIV-uninfected men. Because apparent differences in VAT between HIV-infected and -uninfected persons after matching, adjusting, or stratifying on BMI may be inflated, the FRAM investigators did not directly adjust for BMI in their analyses, noting that "BMI is being influenced by the phenomenon being studied: quantity of fat"[5]. Nevertheless, our findings and those of Joy et al have important implications for the clinician. In both HIV-infected and -uninfected populations, increased VAT is associated with cardiovascular risk factors, such as insulin resistance, low HDL cholesterol, and high triglycerides [20–23], and in the general population higher VAT is associated with incident diabetes mellitus and cardiovascular disease[24, 25]. Clinicians should be aware that some HIV-infected patients, even at a relatively normal BMI, may be at increased risk of adverse metabolic and cardiovascular outcomes attributable to excess VAT.
Our second major finding was that SAT (thigh or abdomen) and extremity fat were markedly lower in HIV-infected men, with or without clinical lipodystrophy, compared to HIV-uninfected controls. After adjustment for BMI or lean body mass, the magnitude and statistical significance of these differences decreased, with differences in thigh SAT between the HIV-infected men without lipodystrophy and the HIV-uninfected group no longer being statistically significant. However, by DXA, extremity fat in HIV-infected men without lipodystrophy was 20–30% lower than in HIV-uninfected men, regardless of the adjustment. This is consistent with the FRAM study, in which HIV-infected subjects with and without clinical lipoatrophy had lower leg fat than HIV-uninfected subjects [5, 6]. Taken together, these results underscore the fact that significant lipoatrophy may be present in HIV-infected persons without clinical evidence of fat wasting and highlight the limitations of using a dichotomous definition of lipoatrophy. Further studies should focus on continuous, objective measures in determining longitudinal changes of body composition in HIV-infected persons and the potential metabolic consequences of mild, subclinical fat wasting.
Few longitudinal data are available on changes in body composition in HIV-infected men with and without clinical lipodystrophy, relative to HIV-uninfected persons. In this study, semi-annual body circumference measurements were available over the 6 years after the substudy visit. HIV-infected men who had clinical evidence of lipodystrophy had a more rapid increase in waist circumference compared to the HIV-infected men without lipodystrophy, and HIV-uninfected men. In contrast, no differences were observed between the groups in the change in hip or thigh circumference over the 6 year interval.
Because measurement of waist circumference does not distinguish between visceral and subcutaneous fat, the more rapid increase in waist circumference in the HIV-infected men with clinical evidence of lipodystrophy could be due to an expansion of either the subcutaneous or visceral fat compartments, or both. Given the severity of lipoatrophy in this group (mean extremity fat 4.5 g), it is possible that some of this increase is due to a reversal of abdominal subcutaneous lipoatrophy. However, if this were the case, more rapid increases in hip and thigh circumference would have also been expected, and these did not occur. Further longitudinal studies, such as the FRAM follow-up study, are required to confirm this finding and understand the extent of change in each of the fat depots in those with a history of body fat abnormalities. Further studies are also needed to understand the factors contributing to the differences in the change of waist circumference in HIV-infected and uninfected patients and whether these are attributable to antiretroviral therapy, increased caloric intake, decreased physical activity, or other factors. In a previous MACS analysis using the entire cohort [8], we found that waist circumference increased more rapidly in HIV-infected men compared to HIV-negative men after adjustment for the effects of antiretroviral therapy, which may suggest a difference in the effect of aging on body composition by HIV-serostatus. Our current findings leave open the possibility that aging-related changes in waist circumference may be accelerated in those with lipodystrophy; this should be further investigated.
The present study had several limitations. First, our cases were defined based on clinical examination alone. In the time since our study was designed, other studies have defined lipodystrophy based on both patient-reported and clinician-observed fat abnormalities[3, 5, 14] which may reduce bias[26]. Second, the MACS population includes only men and our findings are not generalizable to women. Other studies have shown different patterns of fat distribution in HIV-infected men and women compared to gender-matched control populations [5, 6, 16]. In addition, our small sample size may have limited our ability to detect small differences between the groups and precluded analyses based on further stratification of the data, such as comparison of BMI categories. Further studies of body composition comparing HIV-infected subjects and HIV-uninfected controls are required, particularly longitudinal studies to assess changes over time.